lkml.org 
[lkml]   [1998]   [Jun]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: RAID0 poor performances

On a 3 drive stripe on a single ultra-wide controller, with ultra enabled
I see more like 12MB/s

interleaved raid0 array of 3 st34555w 7200 rpm uw scsi disks on bt-958 uw
scsi controller tyan 1668 dual ppro 180 440fx with 128 mb of ram.

Writing the 256 Megabyte file, 'iozone.tmp'...22.650000 seconds
Reading the file...20.980000 seconds

IOZONE performance measurements:
11851455 bytes/second for writing the file
12794826 bytes/second for reading the file



On Sat, 20 Jun 1998, Davide Rossetti wrote:

> On Fri, 19 Jun 1998, Hubert Tonneau wrote:
<SNIP>
> > According to me, 5Mb/s is a resonable performance for a single
> > disk, so 10Mb per seconds for 2 disks sounds fine.
>
> they are four!
>
> > About your test report, i was surprised to see the line:
> > Ultra SCSI: Disabled
> > If you want top performances, you probably need Ultra SCSI bus
> > (20 Mb/S if narrow, 40 Mb if wild)
> > since 2 disks require twice more bandwild on the SCSI bus that
> > a single one.
>
> the theory is: ok 1 disk <= 8 MB/s --> 2 disks could use a 16MB/s
> bandwidth, well under 20 MB/s peak SCSI2-wide bandwidth.
> I use TWO controllers, each one driving TWO disks, so I have a 4 disks
> striping setup.

I'd probably put all four disks in the stripe on the same controller and
turn on ultra scsi since they're st34572s the should take advantage of the
higher signaling rate.

since you're showing slightly higer performance for a single disk then I
was with my st34555w hawk 4xl disks you should be showing incrementaly
more performance from your stripe as well.

> > Ok the Adaptec controler as two SCSI bus: did you verify that
> > the two disks that are linked in each grap are liked to a
> > different SCSI bus.
>
> they are 2 different controller boards !! I you look at the URL I
> referenced, you can see the complete setup. It would be really useful.
>
> Anyway, looking at the md driver source code, that moderate speed-up is ok
> as there is (it seems to me) no I/O overlapping between different disks
> accesses. ie. when a block write operation is dispatched to a disks, it is
> someway waited for completion (ok, not properly, but I tested with
> 250-500 MB write operations, so caching sould have no effect; moreover it
> is a waste of cycles).
>
> regards.
>
> +------------------------------------------------------------------+
> |Rossetti Davide INFN - Sezione Roma I - gruppo V, prog. APEmille|
> | web : http://apemaia.roma1.infn.it/~rossetti |
> | """"" E-mail : davide.rossetti@roma1.infn.it |
> | |o o| phone : (+39)-6-49914412 |
> |--o00O-O00o-- fax : (+39)-6-4957697 |
> | address: Dipartimento di Fisica (V.E.) |
> | Universita' di Roma "La Sapienza" |
> | P.le Aldo Moro,5 I - 00185 Roma - Italy|
> | pgp pub. key: finger rossetti@apemaia.roma1.infn.it |
> | |
> |"Most people think about twice a year. I got famous by thinking |
> | once a week." - George B. Shaw (see Appendix A) |
> +------------------------------------------------------------------+
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
>

--------------------------------------------------------------------------
Joel Jaeggli joelja@darkwing.uoregon.edu
Academic User Services consult@gladstone.uoregon.edu
PGP Key Fingerprint: 1DE9 8FCA 51FB 4195 B42A 9C32 A30D 121E
--------------------------------------------------------------------------
It is clear that the arm of criticism cannot replace the criticism of
arms. Karl Marx -- Introduction to the critique of Hegel's Philosophy of
the right, 1843.




-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu

\
 
 \ /
  Last update: 2005-03-22 13:43    [W:0.577 / U:0.336 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site